Basics about How to write Threats to Validity Section

• Threats to construct validity are concerned with the relationship between theory and observation.
  EXAMPLE: You design some heuristics to build a dataset (e.g., for code reuse) and draw conclusion (about code reuse) on the dataset. But your heuristics may not cover all the code reuse scenarios.
• Threats to internal validity are concerned with the factors that may affect a dependent variable and were not considered in the study.
  EXAMPLE: You consider a paper as a baseline, and the paper does not release its replication package. so you replicate/re-implement the baseline based on your understanding on the paper.
• Threats to conclusion validity are concerned with the relationship between the treatment and the outcome.
  EXAMPLE: You conduct some manual labeling in your work. Does your manual label process can guarantee the correctness of all the cases?
• Threats to external validity are concerned with the generalizability of our results.
  EXAMPLE: Your experiment is conducted in Java, are your findings/conclusions also hold in other PLs?

Another good way to learn how to write is to read the "Threats to Validity" section of the related works.
Please do NOT use "internal threat" to refer to "Threats to internal validity". It's simply WRONG.